Conversation
|
Thanks for creating this PR @chrisb13 ! Sorry just to clarify - which references are you referring to here? |
Thanks @KAUR1984, I'll "@" you in the two code lines now. Please just make the change (as a review suggestion) or commit it directly to this branch. |
|
Ping'ing @AndyHoggANU in case of interest (we've been discussing). |
|
Thanks for your comments @chrisb13 ! Sorry I won’t be able to get back to this until later next week. In the meantime, if anyone else from the @ACCESS-NRI/webops is available to have a look, it would be awesome! |
Thanks for the update @KAUR1984, I think it's a pretty straightforward question however.. How does one do relative rather than absolute links? |
|
Thanks @chrisb13 ! Sorry I wasn’t entirely sure what you meant by fixing the references above. From memory, I believe the link checker in GitHub Actions tends to fail when relative links are used. I just came across a note about this in the Hive Docs README, which seems to confirm it. - styling guidelines:
|
Oh so actually I've done the right thing?! (I thought @atteggiani had a preferred fancier way but I guess I just imagined it.) e.g. this is correct? source: see this comment It won't work in a PR preview then I guess but if that's okay then we can resolve this small query -- thanks! |
|
Hello @chrisb13 , |
Thanks @atteggiani, sounds good but do you mean these comments: I can't see any update from you. Perhaps they are being held and you haven't submitted the review (GitHub does this sometimes I notice until you make a "broad" review comment). |
|
You can see them now. I still had to submit them sorry. |
|
This pull request has been mentioned on ACCESS Hive Community Forum. There might be relevant details there: https://forum.access-hive.org.au/t/forcing-access-om2-using-esm1-5-data/4349/18 |
chrisb13
requested review from
anton-seaice
and removed request for
ezhilsabareesh8
| @@ -586,6 +586,51 @@ env: | |||
| UCX_LOG_LEVEL: 'error' | |||
| ``` | |||
|
|
|||
| #### Using COSIMA restarts with an ACCESS-NRI configuration | |||
| !!! warning | |||
| This workflow is only for advanced users and provided to help users use COSIMA restart files with ACCESS-NRI executables for two key configurations. This workflow will use a mix of restart files: a spun ocean and a sea-ice model that has started almost from rest. This is physically inconsistent and so the model requires an extended spin up and re-evaluation to produce scientifically credible results. | |||
There was a problem hiding this comment.
| This workflow is only for advanced users and provided to help users use COSIMA restart files with ACCESS-NRI executables for two key configurations. This workflow will use a mix of restart files: a spun ocean and a sea-ice model that has started almost from rest. This is physically inconsistent and so the model requires an extended spin up and re-evaluation to produce scientifically credible results. | |
| This workflow is for advanced users to use newer ACCESS-NRI executables with older COSIMA restart files, for two key configurations. This workflow will use a mix of restart files: a spun-up ocean and a sea-ice model that has started almost from rest. This is physically inconsistent and so the model requires an extended spin up and re-evaluation to produce scientifically credible results. |
I think we don't have to caveat needing a spin up for sea ice too much. It needs 5 to 10 years I think but the Andes probably know better
There was a problem hiding this comment.
This works, but is quite klunky. You can do this without needed to do the template run.
Here's my suggested workflow
copy the old restarts (symlink everything except ice if you prefer) :
999 mkdir om2_restarts
1000 cp -r /g/data/cj50/access-om2/raw-output/access-om2-01/01deg_jra55v13_ryf9091/restart795 om2_restarts/
manually make a lice_i2o field, the values are not important and will only be used until the lice field is coupled (probably 1 timestep) :
1001 module load netcdf
1003 module purge
1005 module use /g/data/xp65/public/modules ; module load conda/analysis3
1007 cd om2_restarts/restart795/ice/
1014 ncks -v melt_io i2o.nc lice_i2o.nc
1015 ncrename -v melt_io,licefw_io lice_i2o.nc
1017 ncks -A lice_i2o.nc i2o.nc
repeat for licefh
$ ncrename -v licefw_io,licefh_io lice_i2o.nc
$ ncks -A lice_i2o.nc i2o.nc
then use the copied & modded restart to start the experiment
| !!! warning | ||
| This workflow is only for advanced users and provided to help users use COSIMA restart files with ACCESS-NRI executables for two key configurations. This workflow will use a mix of restart files: a spun ocean and a sea-ice model that has started almost from rest. This is physically inconsistent and so the model requires an extended spin up and re-evaluation to produce scientifically credible results. | ||
|
|
||
| As highlighted above in [start the run from a specific restart file](/models/run-a-model/run-access-om/#specific-restart), there are situations where it is preferable to use an existing restart file rather than start the model from rest, e.g. when the configuration is very expensive to run. New ACCESS-NRI executables may be preferable to use because they contain bug fixes and a newer code base. However, in some cases, if the supplied restart file is not fully compatible with the model configuration, experiments using a custom restart file may require additional manual adjustments to run correctly. |
There was a problem hiding this comment.
This paragraph should go before the sentence starting This workflow will use a mix of restart files
Its also a bit overly wordy and could do with an edit to use the active voice more ?
As highlighted above in start the run from a specific restart file, there are situations where it is preferable to use an existing restart file rather than start the model from rest
could be
It is often preferable to start an experiment from an existing restart file, rather than start from rest, as highlighted above in start the run from a specific restart file.
In general an experiment is realised through a serious of runs, so we should use the word experiment in this case
|
|
||
| Here we provide a workflow for doing these manual adjustments for a popular configuration, namely `release-01deg_jra55_ryf`. | ||
|
|
||
| We first need to identify the restart file we wish to use. For this example we will use `/g/data/cj50/access-om2/raw-output/access-om2-01/01deg_jra55v13_ryf9091/restart795`. It may be useful to consult [this forum thread](https://forum.access-hive.org.au/t/access-om2-control-experiments/258) for making this choice. |
There was a problem hiding this comment.
also maybe consult with the experiment's original author?
| cd release-01deg_jra55_ryf | ||
| payu run | ||
| ``` | ||
| Note: these instructions were tested using commit `83885a45017b3a34ea4fad15d155449c7217ad70`. |
There was a problem hiding this comment.
| Note: these instructions were tested using commit `83885a45017b3a34ea4fad15d155449c7217ad70`. | |
| Note: these instructions were tested using commit `83885a45017b3a34ea4fad15d155449c7217ad70`. |
I am not sure we need this note, but using the config version (2.0) would be easier to understand/consistent with our other messaging:
https://github.com/ACCESS-NRI/access-om2-configs/releases/tag/release-01deg_jra55_ryf-2.0
|
|
||
| Once the above run is complete. We use a mix of restart files. We use the restart file from a sea-ice model that goes almost from rest (the run you just did) and from the COSIMA restart file we take the ocean variables. This does the bulk of the spin up for us. We do this as follows: | ||
| ```bash | ||
| mkdir -p /scratch/tm70/$USER/access-om2/restart_hackom201ryf |
There was a problem hiding this comment.
| mkdir -p /scratch/tm70/$USER/access-om2/restart_hackom201ryf | |
| mkdir -p /g/data/$PROJECT/$USER/access-om2/restart_hackom201ryf |
can we have a clearer name than restart_hackom201ryf
| cd ~/access-om2/release-01deg_jra55_ryf/archive/restart000 | ||
| cp -r ice /scratch/tm70/$USER/access-om2/restart_hackom201ryf |
There was a problem hiding this comment.
| cd ~/access-om2/release-01deg_jra55_ryf/archive/restart000 | |
| cp -r ice /scratch/tm70/$USER/access-om2/restart_hackom201ryf | |
| cp -r ~/access-om2/release-01deg_jra55_ryf/archive/restart000/ice /scratch/tm70/$USER/access-om2/restart_hackom201ryf |
| cp -r accessom2_restart.nml ocean /scratch/tm70/$USER/access-om2/restart_hackom201ryf | ||
| ``` | ||
|
|
||
| We then need to modify the payu configuration to restart from this mix of restart files, there are two steps: |
There was a problem hiding this comment.
| We then need to modify the payu configuration to restart from this mix of restart files, there are two steps: | |
| We then need to modify the experiment configuration to restart from this mix of restart files, there are two steps: |
| cd access-om2 | ||
| payu clone -b expt -B release-01deg_jra55_ryf https://github.com/ACCESS-NRI/access-om2-configs release-01deg_jra55_ryf | ||
| cd release-01deg_jra55_ryf | ||
| payu run |
There was a problem hiding this comment.
I think you could change the runlength so this runs for a day instead of the default length.
| restart: | ||
| /scratch/tm70/$USER/access-om2/restart_hackom201ryf | ||
| ``` | ||
| Finally, submit this new run with `payu run`. |
There was a problem hiding this comment.
I think you can payu sweep --hard before payu run to set the start date back to 1900-01-01
| ``` | ||
| Finally, submit this new run with `payu run`. | ||
|
|
||
| !!! tip |
There was a problem hiding this comment.
The tip should go before "submit this new run" because its not clear what "This" in the tip refers to
|
|
||
| We then need to modify the payu configuration to restart from this mix of restart files, there are two steps: | ||
|
|
||
| 1. We tell CICE that it should not check the date of the restart file, this is because the `accessom2_restart.nml` file in the example above (folder `restart795`) is looking to start at `EXP_CUR_DATE = 2100-01-01T00:00:00` which is not the start date the sea-ice model will have from the one we have just made above in `restart000`, it will have `1900-04-01T00:00:00`. For the first run you'll need to `set use_restart_time = .false.` in `ice/cice_in.nml`. For additional information, see `More complicated cases` in the [COSIMA/access-om2/wiki](https://github.com/COSIMA/access-om2/wiki/Tutorials#updating-an-experiment). |
There was a problem hiding this comment.
You might need to remind people to set use_restart_time = .true. after the first run
|
So these instructions only apply for restarts from experiments which used earlier versions of jra55do. We should note this. Are we confident this fix is applicable to multiple cosima experiments? This content feels more appropriate for a forum post than adding to the main docs website. |
This is a very important point. |
@anton-seaice @chrisb13 |
|
It was discussed, summary here. |
|
|
@chrisb13 Could you please let us know if this PR is ready for review? |
|
It is not, note it being in "draft" currently. |
The much discussed workflow update for using ACCESS-NRI executables with COSIMA created restarts. Thanks to @ezhilsabareesh8 @minghangli-uni for their help and input.
Fixing a bunch of typos and bits that were hard to read.
Few more typo/updates.
Updating the workflow to get around the cice restart date issue.
Trying to fix the numbered list.
Adding handy suggestions from @atteggiani. Co-authored-by: Davide Marchegiani <davide.marchegiani@gmail.com>
atteggiani
force-pushed
the
chrisb/om2_restart_workflow_update
branch
from
04a3adc to
af76214
Compare
atteggiani
force-pushed
the
development
branch
from
254ca6a to
963790e
Compare
7 participants
The much discussed workflow update for using ACCESS-NRI executables with COSIMA created restarts.
Thanks to @ezhilsabareesh8 @minghangli-uni for their help and input.
--
@heidinett @KAUR1984 can you please fix the two references to the earlier section. I wasn't sure how to link to it.